30 Days of AI Engineering — My Content Plan as a Lead AI Engineer

Eleven years into this industry, and I’ll be honest — the last 18 months have been more disruptive to how I work than the previous decade combined. Not in a scary way. In a “I need to document all of this before I forget what it was like before” kind of way.

This is that documentation project.

Over the next 30 days I’m publishing one post a day covering the AI tools and workflows I’m actively using on the job right now: Claude Code for enterprise teams, GitHub Copilot beyond the basics, Microsoft Copilot Studio multi-agent solutions, coding agents, agent skills, AI in the SDLC, and everything in between.

Not tutorials copied from docs. Real stuff, with real friction, real decisions, and real outcomes.

Here’s the full plan.

---

The 30-Day Roadmap

I’ve split this into five thematic arcs so each week builds on the last. Reading them in order makes sense, but each post is also designed to stand on its own.

---

Arc 1 — Foundations & the AI-Augmented Engineer (Days 1–6)

Before diving into any specific tool, I want to set context. These posts are about the mindset, the landscape, and how I actually think about AI in my day-to-day work as a Lead AI Engineer.

Day	Title	What It Covers
1	Why I’m Writing 30 Days of AI Engineering	The “why” behind this series — what’s changed, why now, and what I want to figure out by the end
2	AI in the SDLC — The Honest State of Things in 2026	Where AI genuinely fits in the software development lifecycle today vs. where the hype says it fits
3	The Agentic AI Mental Model Every Engineer Needs	What “agentic” actually means in practice, why it matters, and how it changes the way you design systems
4	Agent Skills 101 — Building Blocks of Useful AI Agents	Anatomy of a well-designed agent skill: inputs, outputs, tool use, memory, and failure modes
5	Choosing Your AI Toolchain — Claude Code, Copilot, or Copilot Studio?	They’re not competing — they solve different problems. Here’s how I think about which to reach for
6	11 Years In, AI-Augmented — How My Workflow Actually Changed	Honest before/after on what AI has changed in how I plan, code, review, and ship

---

Arc 2 — Claude Code for Enterprise (Days 7–13)

This arc goes deep on Claude Code — not the getting-started guide, but the harder questions: how do you use it at scale, in enterprise environments, with real security and governance constraints?

Day	Title	What It Covers
7	Setting Up Claude Code for Enterprise Teams	Auth, security posture, rate limits, audit trails — what you need to figure out before rolling it out
8	Claude Code for Legacy Code Modernization	Using Claude Code to understand, refactor, and document codebases you didn’t write and barely understand
9	Claude CoWork — Async AI Pair Programming in Practice	What Claude CoWork is, how it differs from standard Claude Code usage, and when it’s genuinely useful
10	Prompt Engineering for Coding Tasks — What Actually Works	The prompt patterns I’ve found most reliable for code generation, review, and explanation in real projects
11	Claude Code for Code Reviews at Scale	Using Claude Code as a pre-reviewer: what it catches, what it misses, and how to trust it the right amount
12	Integrating Claude Code into CI/CD Pipelines	Automated code quality, PR summaries, test suggestions — practical integration patterns I’ve tried
13	Claude Code in Regulated Environments — A Security-First Look	Data residency, prompt injection risks, tool access controls — what enterprise security teams will ask you

---

Arc 3 — GitHub Copilot Mastery (Days 14–19)

GitHub Copilot is probably the tool most engineers have tried and feel like they’ve figured out. They haven’t. This arc is about going past autocomplete.

Day	Title	What It Covers
14	GitHub Copilot Beyond Autocomplete — The Features Most People Miss	Copilot Chat, inline fixes, /explain, /tests, slash commands in the terminal — the full picture
15	Writing Better Prompts for GitHub Copilot in VS Code	Comment-driven development, naming conventions, and context tricks that dramatically improve output quality
16	GitHub Copilot for Test Generation — Does It Actually Work?	Honest evaluation: where Copilot-generated tests are good, where they’re dangerous, and how to use them safely
17	Copilot for PR Descriptions, Docs, and Commit Messages	The non-code uses of Copilot that save real time if you know how to set them up
18	Copilot Workspace — Hands-On with Agentic GitHub Copilot	What Copilot Workspace is, how it approaches multi-step tasks, and where it fits vs. Claude Code
19	GitHub Copilot in Enterprise Governance Frameworks	Policy controls, content exclusions, audit logs, and how to get security/compliance teams comfortable with it

---

Arc 4 — Coding Agents & Agent Skills (Days 20–25)

This is the arc I’m most excited about. Coding agents are still early, but they’re already changing how certain classes of engineering work get done. This is where I share what I’ve actually built and learned.

Day	Title	What It Covers
20	What Makes a Good Coding Agent — Design Principles	Scope, tool access, failure recovery, human-in-the-loop checkpoints — the design decisions that matter most
21	Building Your First Coding Agent — A Practical Walkthrough	Hands-on: building a simple but real coding agent, step by step, with the decisions explained
22	Tool Use in AI Agents — Patterns That Work in Production	Function calling, tool chaining, error handling, and the patterns I’ve found reliable vs. brittle
23	Agent Memory and Context Management — The Hard Part	Short-term vs. long-term memory, context window strategies, and how agents lose the thread (and how to prevent it)
24	Multi-Step Coding Agents — Patterns and Failure Modes	What happens when you chain multiple agent actions, where things break, and how to design for recovery
25	Evaluating Coding Agent Quality — Beyond “Did It Run?”	How to actually measure whether an agent is doing the right thing: evals, tracing, human review workflows

---

Arc 5 — Microsoft Copilot Studio & Multi-Agent Solutions (Days 26–30)

The final arc covers the multi-agent work I’m doing right now in Microsoft Copilot Studio — designing, building, and operating agents that work together to solve enterprise-scale problems.

Day	Title	What It Covers
26	Microsoft Copilot Studio for Developers — What You Need to Know	The honest developer’s guide: what Copilot Studio is good at, where it falls short, and how to think about it
27	Designing Multi-Agent Workflows in Copilot Studio	Architecture patterns for multi-agent systems: orchestration vs. choreography, handoff design, shared context
28	Connecting Copilot Studio Agents to Enterprise Systems	Connectors, custom APIs, authentication, and the integration patterns that hold up under real load
29	Multi-Agent Debugging and Observability — Staying Sane	Tracing agent conversations, surfacing failures, and building visibility into systems that are hard to inspect
30	30 Days Done — What I Learned, What Changed, What’s Next	Synthesis post: the biggest surprises, what I’d do differently, and where I think all of this is heading

---

A Few Things I Want to Be Upfront About

This isn’t a sponsored series. I use all of these tools professionally and pay for most of them personally. When something doesn’t work well, I’ll say so.

I’m also not writing for beginners. These posts assume you write code for a living, you’ve at least heard of most of these tools, and you’re past the “should I try AI?” question. You’re in the “how do I actually use this well?” phase. That’s who I’m writing for.

If a post doesn’t land or you want me to go deeper on something, tell me. I’d rather adjust the plan mid-stream than finish 30 days of posts nobody found useful.

Let’s go.

---

Follow along via RSS or bookmark this post — I’ll keep it updated as the series progresses.